current mirror for current steering DAC

[mretsh91]

I am designing a cascode current mirror for current steering DAC, I have a problem in the glitches that occur at switching (shown in the figure shown), Besides, the current value is not accurate

Designed Current Mirror Specs: Iout=220uA
Vbias=0.6 Volt
Rout=5M
Switching Frequency=384MHz



thanks in advance

 

[FvM]

No voltage source or transistor area information in the circuit. Regarding glitches, there's no switching involved in the schematic.

 

[jgk2004]

you are using a voltage source with most likely a 1fs rise/fall time, then yes you get those very large glitches, this is clock feed though and charge injection of turning on and off your switches. Never had your switches being switched by a voltage source, put a real inverter in there so it will be driven with something realistic. Also never use crazy rise and fall times, make it something normal within your technology, for example 90nm to 130m, depending on load and switch transistors you would see rise/fall times like 60 to 100ps

Also, where is your cascode on the mirror part.... both sides should be casdoded, with normal cascode or wide swing by using another diode.

JGK

 

[mretsh91]

I am sorry for the unclear picture. Here in the image attached, there exist two branches where the current is steered between them simultaneously using NMOS switches in SAT region for minimum glitching.


- - - Updated - - -

I am sorry for the unclear picture. Here in the image attached, there exist two branches where the current is steered between them simultaneously using NMOS switches in SAT region for minimum glitching.


- - - Updated - - -

Also, where is your cascode on the mirror part.... both sides should be casdoded, with normal cascode or wide swing by using another diode.

JGK

As I understand, the main role of the cascode on the left side is to provide the DC bias to the the other one. But during design, I found it much easier when removing it and adding an external DC-bias cicruit for this mosfet, Is there a problem in this ?????

 

[jgk2004]

well for starters, I know this is a school project, so if you can use many many voltage sources then sure use them.. But if you are trying to make this as close to a real chip that you could tapeout and measure some day, then you can't have like 10++ pins just for voltage levels....this is why I would practice having one current source (which would come from an on-chip reference, like a gm reference or bandgap), and then just VDD,VSS and maybe a VCM. That is it! That is why you should have your current mirrors correct with cascodes on both sides to match the VDS's of both mirrors.

JGK